Tech Industry Balances Efficiency and AI Opportunities

Morgan Stanley's 2023 Technology, Media and Telecom Conference highlighted trending themes in the sector, with generative AI holding center stage as companies noted that they don’t want to miss out on the potential of this revolutionary “platform shift.” Other trending themes at the event, which drew 1,400 investors and the executives of more than 350 companies, were the need for data observability, the demand for supercomputing and private equity investments in software.

In a market environment characterized by high inflation, rising interest rates and macroeconomic uncertainties, tech investors at the conference said they are prioritizing the companies' near-term profitability while also looking at whether they are growing effectively. For their part, companies are on a mission to prove to investors that they can increase operational efficiency and reduce costs, as blue chips and startups alike have conducted layoffs, cut back on perks or announced stock buybacks to help improve shareholder value.

Generative AI and Assistant Tools

“Generative AI” refers to unsupervised and semi-supervised machine learning algorithms that can use existing text, images, audio or video to create new content. While early results underscore that the models need to be trained, or “tuned,” to produce accurate, trustworthy and valuable outputs, companies point to these algorithms as the basis for future software that will be useful or even essential in every industry and profession.

“AI represents a huge opportunity. It doesn’t just bring greater intelligence, revenue and cost efficiencies to existing infrastructure and applications,” said Brittany Skoda, Global Head of Software Investment Banking. “The re-platforming of industries with AI will generate new use cases that we haven’t even thought possible before.”
 

Corporates and investors at the conference dissected the landscape of generative AI architecture, noting three types of players:
 

• Builders of large language models (LLMs), which let users ask conversational questions and formulate responses based on reams of data
  
• Companies building application programming interfaces (APIs) that allow applications to talk to each other and enterprises to adapt LLMs to their environments
  
• Application companies, which comprise hundreds of startups that want to fine-tune high-value data for enterprises as well as small- and mid-size businesses in specific industries

Supporting this ecosystem are the cloud services providers and semiconductor manufacturers aiming to meet demand for more computing power.

Generative AI has a number of immediate potential applications, which are of interest to investors keen to separate any hype and promises of long-term growth potential from near-term profitability opportunities.
 

• Sales: Salespeople or marketers could quickly search a knowledge base to answer customer questions or automate an assessment of the frequency and sentiment in emails or phone conversations.
• Customer relationship management: Client representatives could use historical data to create forecasts and actionable insights and provide advice. Generative AI applications could take notes, provide summaries, send emails, run a marketing campaign and communicate tailored offerings to upsell. They could also help automate returns, assess customer satisfaction and drive loyalty.
• Ads: Brands could use generative AI to create ad content and reduce the cost of customer acquisition by testing and personalizing ads at scale, thus increasing return on investment.
• E-commerce: Chatbots could offer conversations with customers searching for products, providing human-like questions and suggestions, compared with link results through search engines.
• Human resources: Managers could use generative AI tools to write performance reviews, taking inputs such as remote conversations, and employers could recruit by inputting skill requirements.
• Education: Students could use LLMs as a thought partner to generate baseline arguments, then apply their own critical thinking.
• Content creation: People could use generative AI to build presentations, produce graphic design options and write marketing copy.

For generative AI to flourish in organizations, however, it needs to integrate into their systems, operate in their data and derive insights. At the conference, company executives underscored the near-term business value potential of AI-powered assistants, or interfaces that tap LLMs to create free-flowing conversations. To create effective AI assistants with widespread applications, companies are attempting to harness LLMs, which are working to improve data accuracy, speed and cost.

Investors are analyzing how these tools might accelerate growth and value creation. “Generative AI applications enable companies to increase efficiency and enhance the experience for their customers,” said Diana Doyle, Head of Technology Equity Capital Markets in the Americas. But investors are also considering what it will cost for the creators and users of AI assistants—whether that means hiring people with AI and ML skills to facilitate LLM reinforcement learning and produce secure, accurate and authoritative results, or building or buying generative AI software. 

Data Observability Needs Increase with AI 

As data volume expands with the adoption of public cloud and digital transformation, companies will increasingly require software tools designed for multi-cloud environments to ingest, manage and gain valuable insights from this data. Observability, or the ability to monitor the health of applications and infrastructure in real time, is critical for companies to maintain uptime and optimal performance for their end users and consistent reliability of service. “The cost of downtime is significant,” said Melissa Knox, Global Head of Software Investment Banking. “Digital businesses can lose upwards of $5 million per hour when their applications or infrastructure are down. The focus now is on predicting what will happen and to be able to prevent outages, downtime or poor experiences—and this is achieved through AI.”

Because AI learns with data, the ability to trust the data is important. Companies want to ensure their data remains proprietary so that they can create their own AI models and improve the efficacy of their solutions to improve outcomes for their customers. They also want to prevent data loss and exfiltration and improve overall cybersecurity.

Vendors aim to prove they are the go-to platforms to address observability at scale, which is challenging because data can come from multiple clouds, regions and sources, so effective monitoring should include views into networking, storage, servers and applications. SaaS companies are racing to gain share in the $22 billion observability market and prove their leadership position to investors by adjusting their business models to become consumption based, and their go-to-market functions, enhancing their ability to sell to customers.

AI Accelerates Demand for Supercomputing

The migration of companies’ digital assets, databases and applications from internal infrastructure to the cloud has been decades in the making. But the AI revolution is accelerating the necessity of cloud services, which offer scaling, flexibility and cost savings compared to on-premise infrastructure.

The handful of blue-chip hyperscalers are the largest providers of cloud computing and storage at enterprise scale. But questions around capacity abound because generative AI requires massive amounts of data and computing power to train its models accurately: How much computing capacity is there? How quickly can it run? What is the low-latency capacity? In addition, companies are seeking cloud optimization to reduce costs by right-sizing resources spent on features and determining where to eliminate cloud resource waste.

Scaling super computers also requires chips. LLM builders, enterprises and governments are boosting semiconductor demand, and companies are seeking to improve chips to speed the movement of data in and out of memory with increased power and more efficient memory systems. In addition, electric vehicle (EV) engines and wireless network infrastructure upgrades are driving demand for chips. But customers want to reduce costs, so semiconductor companies are working to design hardware that helps companies reduce the cost of running queries and the retail prices of products such as EVs for end consumers.

One of the biggest challenges for semiconductor companies has been managing supply and demand, as supply chain issues have led to margin costs. Difficulties obtaining equipment and components, plus a surge in demand as companies and consumers prioritized semiconductor-intensive products, led to a chips shortage that became exacerbated during the COVID-19 pandemic. Investors are watching how initiatives to increase domestic production might help high-speed digital applications get the necessary supplies.

Private Equity Investing in Software

Despite investor concerns about macroeconomic headwinds and slowing growth, private equity sponsors are still actively pursuing attractive software opportunities, according to Mike Wyatt, Global Head of Technology M&A. “Growth assets continue to demand premium valuations,” Wyatt said. “PE firms have been active in buying software companies that have recurring revenue and have demonstrated the ability to generate cash flow,” he said, even in a high-interest-rate environment that has made leveraged buyouts more expensive.

The bar for technology initial public offerings continues to remain high. Sponsor buyouts, corporate carveouts and sponsor-to-sponsor trades are seen as more viable alternatives, according to Umi Mehta, Global Head of Technology Private Equity and Venture Capital Investment Banking. “As it stands today, sponsors are effectively at the top of the food chain as active acquirers of technology assets,” Mehta said. Private equity firms continue to be most interested in investing in software assets given their recent performance, financial profile and long-term opportunity, he added.